Cloud Robotics Advances Data in Biological Research
A new effort by Defense Advanced Research Projects Agency (DARPA) has aimed to improve how experiments are planned and designed in multiple scientific disciplines by using computational and analytical tools, including those pulled from engineering disciplines. The Synergistic Discovery and Design Program (SD2) hopes to update protocols with enhanced automation and machine learning strategies. As part of SD2, Ginkgo Bioworks, and Transcriptic have been awarded a dual contract to help advance biological experimentation and iterative testing for synthetic biology. In this blog, we discuss the goals of SD2 and the impact of cloud labs and automation on research with Ben Miles, PhD, Head of Product at Transcriptic .
Ruairi Mackenzie (RM): What are the aims of the SD2 program and how can Transcriptic and Gingko help DARPA achieve them?
Ben Miles (BM): The goal of the SD2 program is to create an approach for conducting research and design in complex and poorly understood systems. This is accomplished by generating huge quantities of experimental data from programmatically addressable labs and feeding these datasets into sophisticated analytical tools. These tools will be used to build system models to facilitate faster design of these complex systems. The project will improve experimental design across disciplines by combining computational, analytical and domain-specific expertise with the generation and integration of terabytes of data. As part of the program, Ginkgo and Transcriptic will use machine learning, closed loop biology, and automated execution of protocols toward the goal of advancing biological experimentation and accelerating iterative testing for synthetic biology.
RM: How can engineering principles help improve the design of biology experiments?
BM: Most biological experimentation can be captured as a multi-parameter optimization process. Whether you are doing medicinal chemistry, organism engineering or assay development, a scientist needs to explore a parameter space efficiently to achieve an objective. There are now many excellent tools for programmatically dealing with a lot of data to navigate this parameter space but the process of generating experimental data remains in the dark ages. By leveraging programmatically driven experimentation, scientists are able to explore more of the parameter space simultaneously and more efficiently find a path to a safe and efficacious drug or another objective.
An additional advancement pioneered by Transcriptic is the development of the open data standard for specifying protocols, Autoprotocol.org. By using Autoprotocol scientists can now codify their protocols as structured objects that can easily be shared, improved, and interpreted by robots or humans. Enabling sharing of protocols between organizations allows easy reproducibility and collaboration on protocol improvements.
RM: I’ve heard of robotic labs, and cloud analytics in labs, but never a robotic cloud lab! Tell us about Transcriptic’s platform and how it will benefit the SD2 program.
BM: Transcriptic’s robotic cloud laboratory platform leverages the cloud paradigm used in the computing industry to allow researchers to conduct more efficient, scalable, and reproducible research, on demand from anywhere in the world. There is amazing talent spread throughout the world today, so one of the goals for SD2 was to have access to labs that could be programmatically used remotely by teams who have powerful analytical tools for building models and hypotheses. The platform is built on the Transcriptic Common Lab Environment (TCLE), a scalable, digital lab operating system that integrates laboratory processes, analytical data, instruments, and IoT technologies into a single platform. As part of the SD2 program, Transcriptic and Gingko’s vast, automated experimental capabilities will be programmatically connected to machine learning-driven design and analysis algorithms. Terabytes of biological data generated by Gingko and Transcriptic will be analyzed using advanced artificial intelligence to design new experiments to be performed on the 24/7 automated experimental labs and the cycle will continue
RM: Does widescale automation threaten jobs in research in the same way as it has threatened jobs in other industries?
BM: No, we do not expect our robotic cloud lab to threaten jobs. Instead it will allow trained researchers to focus on the science versus mundane, tedious tasks. In addition, our platform offers seamless data integration and sharing, which encourages global collaboration. This benefit has already been implemented in the SD2 program and is proving to be very advantageous to the participants in generating sophisticated analytics pipelines. Companies are using our robotic cloud lab to conduct the labor-intensive work of scientific experimentation, allowing researchers to focus on generating hypotheses and analyzing data. Pharmaceutical and biotech companies are looking for ways to conduct large volumes of research work more efficiently, and many companies are already outsourcing work to third party CROs. Transcriptic and cloud labs can really be seen as an evolution of this outsourcing, and our ability to enhance reproducibility has the potential to more rapidly advance science. Transcriptic essentially provides all the benefits of outsourcing, with tighter integration into a company’s research infrastructure and greater control for the scientist.
Ben Miles was speaking to Ruairi J Mackenzie, Science Writer for Technology Networks